<![CDATA[The development of artificial intelligence (AI) has influenced how society accesses and understands religious information, including Qur’anic interpretation studies. ChatGPT is one of the technologies used to support the understanding of religious texts, but its ability to handle prompt manipulation in interpreting mutashabihat verses requires further study. This research aims to analyze the influence of prompt variations on ChatGPT responses in interpreting QS. Thaha: 5 and evaluate its consistency based on tafsir standards. This study uses a qualitative library research approach with experimental testing through three types of prompts: neutral, persona, and manipulative. ChatGPT outputs were analyzed using content analysis and compared with Tafsir Ibn Kathir and Tafsir Al-Misbah. The results show that ChatGPT maintains interpretation boundaries according to tafsir parameters, including when facing manipulative prompts that lead to incorrect theological interpretations. However, the system has limitations because it does not possess religious understanding and moral judgment. Its responses are generated from previously learned textual data and language patterns, not from belief or religious understanding. This study shows that AI can support tafsir studies but still requires human supervision and verification.]]>
Copyrights © 2026
